racadm fwupdate fails

[Harold van Oostrom]

Hello Bruce,

This may be unrelated but I've observed the following on all of 
our PE1650's having the same packages and redhat version that you 
mention and also on systems having the newer OMSA-1.2.2a with
dellomsa-4.71-3652 and dellomsa-drivers-4.71-3652

The systems are configured to start only dellomsa on boot i.e.
racser racvnc racsrvc and omawsd are all `/sbin/service off -ed'

After booting the system, the RAC seems accesible through racadm:
 ~# racadm getsysinfo

RAC Information:
RAC Date/Time      = Fri, 27 Dec 2002 00:16:58 GMT+01:00
Firmware Version   = 1.10 (Build 14 1)
Firmware Updated   = 
Hardware Version   = A00
Current IP Address = 192.168.1.123
Current IP Gateway = 192.168.1.64
Current IP Netmask = 255.255.255.0
DHCP enabled       = FALSE
PCMCIA Card Info   = N/A

etcetera.

Then after some time, haven't investigated how long exactly but mostly I
looked at it after a day or so, all of a sudden exactly the same command 
doesn't work anymore:

 ~# racadm getsysinfo
racadm getsysinfo:: transport failed

This happened even on systems that are just powered on without users or 
network and only regular cronjobs.

On at least two occasions I have used the following procedure with success
(with a success rate of about one in five) to restore racadm operation to 
normal:

/etc/init.d/dellomsa stop
/etc/init.d/snmpd stop
rmmod esm
(b)reset the RAC board (bet you didn't know you could telnet into it)
sleep 60
/etc/init.d/dellomsa start

 ~# racadm getsysinfo

RAC Information:
RAC Date/Time      = DSU+00:05:54
Firmware Version   = 1.10 (Build 14 1)  (*note below)
...
etc.

I know a sure way to make racadm fail too:
/etc/init.d/omawsd start

I.e. start the OMSA webserver listening on port 1311 ..
after which again racadm commands fail with the same

 ~# racadm getsysinfo
racadm getsysinfo:: transport failed

In addition we have seen these errors

racadm config: Transport: ERROR: objectIdSet. RC=15

racadm getconfig: Transport: ERROR: objectIdGet. RC=14
Segmentation fault

Which could be cured by disconnecting the server from power.
We have the ERA/O, O for on-board.